TL;DR
This paper presents a comprehensive, harmonized dataset of forest biomass and increment across 38 European countries, offering unprecedented spatial resolution and consistency. This dataset is crucial for sustainable forest management, bioeconomy planning, and climate change mitigation efforts by providing a unified view of Europe's forest resources.
What Problem Does This Paper Address?
Accurate and consistent data on forest biomass and its growth (increment) are vital for sustainable forest management, policy-making, and understanding Europe's role in the green transition and climate neutrality. However, existing datasets suffer from a lack of harmonization in definitions, methodologies, and temporal coverage across different European countries. National Forest Inventories (NFIs), while detailed, use country-specific approaches, making direct comparisons difficult. This hinders a unified European-level assessment of forest resources, their availability for wood supply, and their contribution to carbon sequestration. This paper aims to bridge this gap by creating a harmonized, high-resolution dataset that overcomes these limitations.
Methods
The core of this work lies in the extensive harmonization of data from various sources. National Forest Inventory (NFI) data from 26 countries, representing 92% of Europe's forest area, were harmonized using common definitions for biomass pools and estimation methods. This harmonized data was then updated to a common reference year (2020 for biomass, 2015 for increment) using the Carbon Budget Model (CBM) or linear adjustment functions. For countries lacking harmonized NFI data, statistics were derived from the State of Europe's Forests (SoEF) 2020 report.
To create high-resolution maps, the harmonized statistics were used to calibrate and bias-correct existing satellite-derived maps (e.g., ESA CCI biomass map, Copernicus Forest Type map). This involved several steps: creating a forest mask consistent with the statistics, adjusting areas with no biomass, and applying a bias-correction factor to align map values with sub-national statistics. This ensures that the maps spatially represent the harmonized statistical data at a 100m resolution.
Datasets
The primary datasets used are:
- National Forest Inventory (NFI) data: Collected from 26 European countries, forming the basis for harmonized statistics.
- State of Europe's Forests (SoEF) 2020 Report: Used for countries lacking harmonized NFI data or for specific variables.
- Carbon Budget Model (CBM): A forest growth model used for temporal harmonization of biomass and increment data.
- Copernicus Forest Type map (2018): Used to create a forest mask for map adjustments.
- ESA CCI Biomass map (2020): Used as a base for creating the harmonized biomass maps.
- European Digital Elevation Model (EU-DEM): Used for mapping slope restrictions.
- World Database on Protected Areas (WDPA): Used for mapping protected areas.
- JRC European Atlas of Forest Tree Species: Used for mapping protected species.
- OpenStreetMap: Used for mapping road networks to assess accessibility.
- MODIS data (MYD13Q1 V6.1): Used to compute the kNDVI vegetation index for mapping productivity.
The paper introduces a new, harmonized European-wide dataset of forest area, biomass stock, forest available for wood supply (FAWS), and forest increment, presented at both sub-national and national scales, with accompanying 100m resolution maps.
Key Insights and Ideas
- The critical need for harmonization: The paper strongly emphasizes how differing national definitions and methodologies create significant comparability issues, hindering pan-European analysis. Their harmonization process directly addresses this.
- Bridging statistics and maps: A key innovation is the explicit use of harmonized statistics to bias-correct and align satellite-derived maps, ensuring spatial representations are consistent with ground-based inventory data. This is crucial for applications requiring both detailed spatial information and statistical rigor.
- Uncertainty as a feature: The dataset includes an uncertainty indicator, allowing users to filter data based on reliability, acknowledging that data derived from different sources (NFI vs. SoEF vs. gap-filled) have varying levels of confidence.
- Forest available for wood supply (FAWS) is a complex metric: The paper highlights that FAWS is not just about forest area but also about numerous restrictions (slope, altitude, protection, accessibility, productivity), which are mapped and accounted for.
Key Results
- Comprehensive coverage: The dataset includes statistics and maps for 38 European countries.
- High spatial resolution: Maps are provided at a 100m resolution.
- Harmonized statistics: For 26 countries (92% of Europe's forest area), biomass statistics are harmonized for definition and updated to 2020. Increment statistics are harmonized for 10 countries (71% of forest area) and updated to 2015.
- Improved map accuracy: Bias-adjusted forest area maps show a relative Mean Absolute Error (rMAE) of only 0.3% compared to reference statistics, a significant improvement over the original Copernicus map (10.8%). Similarly, the bias-adjusted biomass map achieved an rMAE of 0.3% compared to reference statistics, down from 18.2%.
- Temporal harmonization impact: Temporal harmonization to 2020 for biomass increased the total biomass stock of the 26 countries by 11.3%, demonstrating the importance of updating older NFI data.
Limitations and Future Work
- Data source variability: For countries not included in the harmonized NFI data, statistics are derived from SoEF or CBM, which may have lower comparability and no explicit uncertainty information.
- Gap-filling uncertainty: Missing data for increment and FAWS were estimated using adjustment factors or data from neighboring countries, introducing higher uncertainty.
- Map vs. Statistics Discrepancies: While maps are bias-adjusted to match statistics, random errors remain at the pixel level. Furthermore, FAWS maps and statistics may not always identify the exact same restrictions, making them complementary rather than identical.
- Future work: The authors suggest that future improvements could involve using novel sensor data (e.g., lidar, P-band radar) for higher pixel-level accuracy in biomass mapping. Refining FAWS maps with more detailed national restriction data and integrating them more closely with statistics is also a potential avenue.
Summary updated